PROJECT SUMMARY/ABSTRACT Genetically-engineered mouse models (GEMMs) are essential tools for the study of cancer. However, there is growing concern that GEMMs fail to recapitulate the mutation burden of human carcinomas. GEMMs have startlingly low overall mutation rates, far below what is observed in their human counterparts. This makes such models useful for studies of oncogenic signaling pathways, but greatly restricts their utility for studies of genetic heterogeneity and clonal variation, tumor immunology, or the impact of mutational load/base substitution rates on tumor behavior and response to therapy. The latter has become particularly relevant with the advent of immune checkpoint therapies, given that the best predictor of treatment success is a high incidence of somatic mutations, irrespective of tumor type. The same limitations are likely to be encountered with GEMMs based on newer genome-editing methods, pointing to the need for alternative approaches to optimize with respect to mutational load, which we now know defines so many aspects of tumor biology, clinical behavior and treatment response. In this project, submitted in response to PAR-17-245 ?Research Projects to Enhance Applicability of Mammalian Models for Translational Research?, we propose to generate and characterize the first mouse cancer models based on polymerase-driven ultramutation. These approaches will 1) catalyze modelling of any cancer driven by POLE ultramutagenesis and 2) permit efficient ?humanization? of any GEMM with respect to mutational load. Our approach represents a new and widely-applicable route to the creation of mouse models that recapitulate the mutational loads inherent to human cancer. These new genetic tools and the diverse animal models they will enable will stimulate a wide range of translational and preclinical investigations for which GEMMs were previously not well-suited, thus fulfilling the goals of PAR-17-245.